def __init__(self, content, index=0, name=None, score=0):
"""Accepts either a filepath or a list of words, possibly with ranks."""
if isinstance(content, str):
words = list(read_wordlist(content, score))
self.path = content
else:
self.path = None
words = [(w if isinstance(w, tuple) else (w, score)) for w in content]
# for now, concat compound words
words = [(w.replace(" ", ''), score) for w, score in words]
# reject all non-alphabet chars
ord_A = ord("A")
ord_Z = ord("Z")
ord_a = ord("a")
ord_z = ord("z")
def is_ok(w):
for c in w:
ord_c = ord(c)
if not (ord_A <= ord_c <= ord_Z
or ord_a <= ord_c <= ord_z):
return False
return True
words = [item for item in words if is_ok(item[0])]
self.words = cPalabra.preprocess(words, index)
# keys of self.words = lengths
# values = list of words of that length with (word, score)
self.index = index
self.name = name
def remove_words(self, words):
"""
Remove a list of words from the word list.
This method removes the words and rebuilds the word list.
"""
rebuild = False
for item in words:
key = len(item[0])
if item in self.words[key]:
self.words[key].remove(item)
rebuild = True
if rebuild:
new_words = []
for l in self.words.keys():
new_words.extend(self.words[l])
self.words = cPalabra.preprocess(new_words, self.index)